Re: [FFmpeg-devel] [PATCH] avfilter/f_ebur128: multiply in integer first, before dividing in float

2022-04-06 Thread Paul B Mahol
ok, if other fates are not regressed.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".


[FFmpeg-devel] [PATCH] avfilter/f_ebur128: multiply in integer first, before dividing in float

2022-04-05 Thread Hendrik Leppkes
Restores the order of operations from before 15a1104, which reduces
errors from floating point calculations, and fixes FATE on mingw64.
---
 libavfilter/f_ebur128.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libavfilter/f_ebur128.c b/libavfilter/f_ebur128.c
index 1b1bcf1bff..a036aa0c86 100644
--- a/libavfilter/f_ebur128.c
+++ b/libavfilter/f_ebur128.c
@@ -768,7 +768,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame 
*insamples)
 
 /* get lower loudness to consider */
 n = 0;
-nb_pow = LRA_LOWER_PRC * 0.01 * nb_powers + 0.5;
+nb_pow = LRA_LOWER_PRC * nb_powers * 0.01 + 0.5;
 for (i = gate_hist_pos; i < HIST_SIZE; i++) {
 n += ebur128->i3000.histogram[i].count;
 if (n >= nb_pow) {
@@ -779,7 +779,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame 
*insamples)
 
 /* get higher loudness to consider */
 n = nb_powers;
-nb_pow = LRA_HIGHER_PRC * 0.01 * nb_powers + 0.5;
+nb_pow = LRA_HIGHER_PRC * nb_powers * 0.01 + 0.5;
 for (i = HIST_SIZE - 1; i >= 0; i--) {
 n -= FFMIN(n, ebur128->i3000.histogram[i].count);
 if (n < nb_pow) {
-- 
2.33.0.windows.2

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".